Model reference · open weights

ai-light-dance_drums_ft_pretrain_wav2vec2-new_onset-idmt-mdb-2

Name: ai-light-dance_drums_ft_pretrain_wav2vec2-new_onset-idmt-mdb-2
Brand: AxForge

Available as managed deployment Audio gary109 · community Speech→text 1 variants 1k dl/mo

ai-light-dance_drums_ft_pretrain_wav2vec2-new_onset-idmt-mdb-2 is an open-weight audio or speech model from gary109. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.

Request deployment All audio & music

Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.

What it is

Released by	gary109
Type	Audio & music
Task	Speech→text
Runs with	transformers
Released	2022-11-25
Popularity	1k downloads / month
Licence	Unknown

About

What ai-light-dance_drums_ft_pretrain_wav2vec2-new_onset-idmt-mdb-2 is

This model is a fine-tuned version of gary109/ai-light-dance_drums_ft_pretrain_wav2vec2-base-new_onset-idmt-mdb-2 on the GARY109/AI_LIGHT_DANCE - ONSET-IDMT-MDB-2 dataset. It achieves the following results on the evaluation set:

Loss: 0.4204
Wer: 0.1844

Read the full model card

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0003
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 30
num_epochs: 100.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2599	0.98	11	0.4281	0.2198
0.2491	1.98	22	0.4891	0.1947
0.2619	2.98	33	0.5496	0.2183
0.3354	3.98	44	0.5202	0.2094
0.277	4.98	55	0.4574	0.2080
0.3065	5.98	66	0.4749	0.2080
0.2669	6.98	77	0.5902	0.2183
0.2829	7.98	88	0.8560	0.2050
0.2509	8.98	99	0.6190	0.2035
0.2728	9.98	110	0.6562	0.2109
0.2615	10.98	121	0.6291	0.2065
0.2586	11.98	132	0.6167	0.1844
0.2441	12.98	143	0.6736	0.1962
0.233	13.98	154	0.5727	0.2050
0.2567	14.98	165	0.6165	0.1873
0.2264	15.98	176	0.7506	0.2080
0.2346	16.98	187	0.7017	0.1888
0.2343	17.98	198	0.5930	0.2094
0.2638	18.98	209	0.5730	0.2006
0.2543	19.98	220	0.4991	0.2198
0.2476	20.98	231	0.6364	0.2065
0.2777	21.98	242	0.6247	0.1844
0.2661	22.98	253	0.5589	0.2006
0.2094	23.98	264	0.5316	0.2080
0.2496	24.98	275	0.8821	0.1844
0.2302	25.98	286	0.5408	0.1814
0.2651	26.98	297	0.6479	0.2094
0.2119	27.98	308	0.5875	0.1814
0.2468	28.98	319	0.7614	0.1976
0.2239	29.98	330	0.4908	0.1903
0.2514	30.98	341	0.5196	0.2035
0.2244	31.98	352	0.5580	0.1991
0.2524	32.98	363	0.5342	0.2021
0.2516	33.98	374	0.4204	0.1844
0.2515	34.98	385	0.5135	0.2124
0.2542	35.98	396	0.8150	0.1962
0.2269	36.98	407	0.8833	0.2094
0.212	37.98	418	1.3235	0.2183
0.2119	38.98	429	0.6919	0.2021
0.2228	39.98	440	0.6712	0.2021
0.2127	40.98	451	0.7557	0.1976
0.2064	41.98	462	0.5918	0.1947
0.2147	42.98	473	0.8049	0.1962
0.193	43.98	484	0.7117	0.1976
0.2063	44.98	495	0.5544	0.1962
0.1989	45.98	506	0.5782	0.1888
0.2193	46.98	517	0.5216	0.1947
0.2012	47.98	528	0.5269	0.1917
0.2187	48.98	539	0.4636	0.1844
0.2128	49.98	550	0.4968	0.1888
0.2041	50.98	561	0.4784	0.1888
0.1993	51.98	572	0.5592	0.1755
0.1981	52.98	583	0.4871	0.1785
0.1808	53.98	594	0.4771	0.1740
0.2317	54.98	605	0.5285	0.1814
0.1906	55.98	616	0.5485	0.1844
0.1924	56.98	627	0.5615	0.1814
0.1761	57.98	638	0.4604	0.1799
0.2047	58.98	649	0.4223	0.1829
0.1992	59.98	660	0.4706	0.1873
0.1949	60.98	671	0.4633	0.1844
0.2034	61.98	682	0.4854	0.1814
0.2147	62.98	693	0.4489	0.1844
0.2135	63.98	704	0.4874	0.1726
0.2021	64.98	715	0.4635	0.1814
0.1822	65.98	726	0.4813	0.1785
0.1882	66.98	737	0.5076	0.1799
0.2014	67.98	748	0.5183	0.1888
0.1869	68.98	759	0.5035	0.1799
0.1914	69.98	770	0.4694	0.1844
0.1972	70.98	781	0.4485	0.1844
0.1724	71.98	792	0.4579	0.1829
0.195	72.98	803	0.5178	0.1814
0.2017	73.98	814	0.4978	0.1829
0.1874	74.98	825	0.5035	0.1873
0.1925	75.98	836	0.5495	0.1829
0.1845	76.98	847	0.5394	0.1799
0.1718	77.98	858	0.5070	0.1711
0.1824	78.98	869	0.4912	0.1770
0.1702	79.98	880	0.4632	0.1726
0.1563	80.98	891	0.4412	0.1726
0.1858	81.98	902	0.4635

From the published model card. Full card on the HuggingFace links in the sidebar.

Using it via the API

Call it like any OpenAI endpoint

Once AxForge deploys ai-light-dance-drums-ft-pretrain-wav2vec2-new-onset-idmt-mdb-2 for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (ai-light-dance-drums-ft-pretrain-wav2vec2-new-onset-idmt-mdb-2 below is illustrative; you get the exact model name on deployment.)

$ curl -sS https://api.axforge.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -F model="ai-light-dance-drums-ft-pretrain-wav2vec2-new-onset-idmt-mdb-2" -F file=@audio.mp3

Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.

Get API access

Sizes, precisions & builds

ai-light-dance_drums_ft_pretrain_wav2vec2-base-new_onset-idmt-mdb-2

— · BF16

Weights ↗

Details

Trained / evaluated on

ai_light_dance